- Market processes are not visible. For instance, when a government taxes its citizens and offers a subsidy to some producers, what is seen is the money taken and the money received. What is unseen is the amount of production that would occur in the absence of such transfers.
- Markets are intrinsically probabilistic and therefore marked with uncertainty, like other living organisms, we are loss-averse and try to minimise uncertainty
- Humans may be motivated to place their trust in processes that are (or at least seem to be) driven by agents rather than impersonal factors.
The last point reminded me of speculation from the recent LessWrong article Conspiracy Theories as Agency Fictions:
Do all theories of legitimacy also perhaps rest on the same cognitive failings that conspiracy theories do? The difference between a shadowy cabal we need to get rid of and an institution worthy of respect may be just some bad luck.
Before thinking about these points and debating them I strongly recommend you read the full article.
http://www.economist.com/node/21525456 for the starter.
The arguments in favour of speculation, in general, rest on assumption of intelligent trading and diversity of strategies. The short term range, however, is entirely up to software tools, whose decisions are stupid and have identical systematic errors. I think the short term trading is going to go out with a bang first time there's any interesting software-level exploitation - either a direct hack or adoption of one bot whose decision theory makes non-trivial use of understanding of other instances of itself. Bang as in, a millisecond level bubbles and crashes followed by decision to roll it back and regulate or prohibit.
Wow, sorry about the stupid autocorrects above.
As have been pointed out on e.g marginalrevolution.com, one reason HFT is so popular is because the minimum stock price increment is one cent. HFT might conceivably lose much of its allure if this lower bound is changed to, say, .01 cent.
I'm not convinced trading must be intelligent to provide beneficial information to the market. I'm also not convinced all HFT systems have identical systematic errors. Can you give some examples?